The Eye of Autonomous Driving - Introduction to Camera Modules
May 16, 2022
The Eye of Autonomous Driving - Introduction to Camera Modules
The importance of cameras to ADAS cannot be overstated. Recently, the interface design of the domain control camera module is in progress. It is found that the information and descriptions on the Internet are of different calibers. Master data transmission and interface definitions, and provide reference for camera module selection and interface design of automotive domain controllers.
As shown in the figure below, the lens (Lens) + image sensor (Image Sensor) + image signal processor (Image Signal Processor, ISP) + serializer (Serializer) is the most basic structural block diagram of the camera. The general step is to collect the basic information of the object through the lens, and then processed by the Image Sensor (hereinafter referred to as the Sensor), and then handed over to the ISP for processing, and then used in the coaxial cable or used for GMSL (Gigabit Multimedia Serial Link) to transmit dual Stranded. Below is a description of each extension, including data transfer.
1. Lens
The poor play with cars, the rich play with watches, the poor have three generations of photography, and SLRs have ruined their lives. Everyone must have heard that the lens here refers to the expensive lens in photography. It will not be described in detail here, nor is it a parameter related to our controller design. Let's talk about a more intuitive parameter xGyP, where x and y refer to numbers, G refers to Glass lens, and P refers to Plastic. In general, the larger the number of x, the more expensive the lens.
2. Image sensor
Sensors use photodiodes to convert light and electricity into digital information. At present, it is generally divided into two types of image sensing elements: CMOS (Complementary Metal Oxide Semiconductor) and CCD (Charge Coupled Device). Since each pixel of the Sensor can only be sensitive to R, G or B light, each pixel stores monochromatic light, which is RAW DATA (same as RGB RAW DATA). If the arrangement format of this raw data is RGRG/GBGB arrangement, we call it Bayer RGB (this is the most common). Therefore, Bayer RGB belongs to RAW DATA, but RAW data is not necessarily Bayer RGB. Sensors from different manufacturers have different RAW DATA arrangements.
2.1 CCD
The CCD sensor is shown in the figure below. The charge signal of each pixel (Pixel) in each row will be transmitted to the next pixel in turn, output from the bottom, and then amplified and output by the amplifier on the edge of the sensor. It is to add A/D conversion after the bus.
2.2 CMOS
The CMOS sensor is shown in the figure below. Each pixel is connected to an amplifier and analog-to-digital conversion circuit, and outputs a signal in a manner similar to a memory circuit. Typically, A/D conversion is added next to each photodiode.
2.3 The difference between the two
1. Read information differently
The CCD has only one bus, and passively outputs the collected data under the control of the synchronization circuit. The output data is the level value of the corresponding diode which is gradually shifted in synchronization with the clock signal. Charge information transmission and read output require a clock control circuit and three sets of different power supplies, and the entire circuit is more complicated.
CMOS has two buses, actively outputs the collected data information, directly reads the level of the bus in the form of coordinates (transistor switch array), and saves the level value of each pixel.
2, the speed is different
It can also be seen from the way of reading information that the CCD needs to output information bit by bit in units of "lines" under the control of the clock pulse, and the speed is relatively slow.
CMOS has an array of transistor switches. The Sensor takes out the electrical signal while collecting the optical signal, and can also process the image information of each unit at the same time. It can run at relatively high frame rates. For example, some CMOS designed for machine vision claim frame rates as high as 1000 frames per second.
3. Different power and power consumption
CCD CCD sensors are passive capture and require an external voltage to move the charge in each pixel, typically 12 to 18V. Generally, one clocked power supply and three sets of power supplies are required, and the high drive voltage makes it consume more power. much higher than CMOS sensors.
The image capture method of the CMOS photoelectric sensor is active, and the charge generated by the photodiode will be directly amplified and output by the transistor next to it. Usually only 3V or 5V power supply is required, and the power consumption is very small, only 1 CCD charge coupler is required. /8 to 1/10.
4. Different picture quality
CCD charge-coupled device manufacturing technology started earlier and the technology is mature. It uses a PN junction or a silicon dioxide (SiO2) isolation layer to isolate noise. With only one amplifier at the edge of the CCD sensor, the noise is low and the image quality is better than CMOS.
CMOS photoelectric sensors have high integration. The distance between each photoelectric sensing element and the circuit is very close, and the optical, electrical and magnetic interference between them is serious, and the noise has a great impact on the image quality. In recent years, with the development of CMOS circuit noise reduction technology, the continuous progress of CMOS imaging devices has provided good conditions for the production of high-density, high-quality CMOS imaging devices, and the imaging quality has been significantly improved. The pixel size of CMOS is difficult to reach the level of CCD sensor. Comparing CCD and CMOS of the same size, the resolution of CCD is usually better than that of CMOS sensor.